decimal问题:计算结果和精算表结果对比发现结果5舍6入

解决办法:把udaf自定义函数中的decimal位数增大一点即可

 @F.pandas_udf("decimal(12,10)")
    def udf_lx(qx:pd.Series,lx:pd.Series) -> decimal:
        decimal.getcontext().rounding = "ROUND_HALF_DOWN"
        temp1 = Decimal(0)
        temp2 = Decimal(0)
        for i in range(0,len(qx)):
            if i == 0:
                temp1 = Decimal(lx[i])
                temp2 = Decimal(qx[i])
            else:
                temp1 = (temp1 * (1-temp2)).quantize(Decimal('0.0000000000'))
                temp2 = qx[i]

        return temp1

    spark.udf.register('udf_lx',udf_lx)